A cost analysis of the first year after stroke – early triage and inpatient rehabilitation may reduce long term costs

Aim of the study: To analyse the costs of stroke in the first year covered by insurance companies and to correlate them with the clinical outcome data. Methods: We contacted the insurance companies of 172 consecutive stroke patients of a single institution cohort for a detailed report of the stroke costs. A complete data set over one year was obtained from 131 patients (76%). Results: Severity of stroke was significantly associated with increasing total costs (p = 0.0002). The rehabilitation clinic made up 37% of the total costs followed by nursing home with 21% and acute hospital with 21%. Mean cost of stroke per patient was 31,115 CHF in the first year. Costs per patient for inpatient rehabilitation were similar to those for the nursing home after one year; however, the Barthel-index of patients with inpatient rehabilitation increased by 42 ± 29 points as compared to patients without inpatient rehabilitation by 23 ± 26 points (p <0.05), and 86% resp. 81% of patients with inpatient stroke rehabilitation lived independently after 6 and 12 months respectively. Conclusions: The high level of independence after inpatient stroke rehabilitation underlines the importance of patient selection and/or rehabilitation. Therefore, long-term stroke costs may be significantly reduced by an early and careful triage in the case management after stroke and a case-dependent investment in initial costly appearing inpatient rehabilitation.


Introduction Patients and methods
A prospective cohort study of a single institution (KSB) evaluated the quality of stroke treatment and outcome in 172 consecutive patients from 1/2002 to 3/2003.In particular, the health status before the event, stroke risk factors, acute event (size, location, clinical presentation), the psychosocial consequences, and quality of life in the subsequent 6 and 12 months were analysed systematically [2].The study protocol was approved by the Institu-tional Review Board of the hospital.Informed consent was obtained from all patients or their first relatives if patients were unable to give it.
The present cost analysis is a subgroup of 131 patients from this study showing exactly the same outcome data for the total population.Included were patients with the signs and symptoms of an acute stroke according to the WHO definition of acute stroke.Patients had to have been living at home before the event.In order to allow a direct comparison with the Swiss cohort of the international stroke trial (IST) -patients with thrombolytic therapy and/or need for neurosurgical intervention were excluded.Further, TIA with full recovery within 24 hours after the event was an exclusion criterion.All patients received brain imaging (CT or MRI) and were evaluated for their cardiovascular risk factors, medications and received an extensive workup including standard laboratory evaluation, ECG, 24h-ECG trans-thoracic and transoesophageal, echocardiography, doppler ultrasound of the cerebral arteries.Disease severity was analysed prospectively immediately by our staff and within 24 hrs (= time point B1) of the hospitalisation by a neurologist; the same neurologist (W.J.) re-examined the patient again after 3 (= time point B2) and 6 (= time point C) months.Stroke severity was quantified by stroke scales including the NIH stroke scales, the Barthel-indices and the Rankingscores at the times indicated above.Quality of life was measured by the SF-36 questionnaire.The clinical diagnosis of "dependency" was defined as the inability to manage the daily activities such as washing, dressing, eat-   The detailed cost-analysis started at the day of the incident stroke and lasted for 1 year.We obtained complete data sets from 131/172 patients (76%); among them were all the major Swiss insurance companies.One company went bankrupt in the meantime (6 patients), others were unwilling or unable to transfer their data (33 pts).No selection of particularly expensive or of low cost patients was evident.80 patients had basic insurance and 51 had additional private insurance coverage.We calculated the costs both ways and the costs of the private insurances are included in the results as well.
Criterion for inpatient stroke rehabilitation was a combined prognostic analysis by the neurologist comprising the size and location of the cerebral lesion, the severity of impairment (Barthel-Index, NIHSS), the patient's co-morbidity and age.The time of discharge from inpatient rehabilitation was discussed in weekly interdisciplinary team reports depending on the individual progress during rehabilitation.Outpatient clinic services consisted of all services including physiotherapy, ergotherapy, and speech training.
We divided the costs into the following categories: Prespecified subgroup analysis included a) the costs of subsequently independent vs dependent patients one year after the stroke event, b) the costs of patients who underwent inpatient rehabilitation vs. the ones who did not (ie, were admitted to the nursing home or discharged home, figure 1).
Furthermore, a rank order of the expenses per patient was established and compared with outcome.

Statistical Analysis
Fisher´s exact test was used where appropriate.The total costs were statistically further analysed in a multiple regression model where patient characteristics were included if they contributed significantly.Thus possible bias effects were investigated.We included age, gender, Rankingscore, Barthel-index and NIHSS (within 24 hrs [= B1] and 6 months after stroke [= C]), inpatient rehabilitation, atrial fibrillation and the loss of consciousness in the model, which included all surviving patients.Because of the skewed distribution, logarithmic costs were considered in the linear regression model.The assumptions of this model are holding well.A backward selection procedure with level 5% determined a significant model which includes inpatient rehabilitation, Rankingscore at time point C, NIHSS at time point B1, and also a small interaction between inpatient rehabilitation and NIHSS at time point B1.
Inpatient rehabilitation had the largest impact with a regression coefficient 0.955 (p <0.001) on the mean logcosts.Rankingscore at time point C has a linear increasing effect on the log-costs mean with regression coefficient 0.175 (p = 0.0002).The impact of significant effects of other categories of intervention was much smaller.Obviously some of the indices are correlated.So, instead of NIHSS and Rankingscore other indices (such as Barthelindex) could have been selected with a resulting model almost as good as the reported one.
Out of these 131 patients with complete financial data sets over 12 months, 44% underwent inpatient rehabilitation, and 12% (13%) had died at 6 and 12 months, respectively.75% of the survivors were independent after one year.Initial stroke severity was 7 ± 6 by the NIHSS.The patient characteristics are summarised in table 1 and the outcome data in table 3.
The mean costs per patient were CHF 31,115.-±23,752.-SD(from 3,256.-to122,443.-,median 26,370.-).The wide range between minimal and maximal costs is reflected by the spectre of the disease severity: 6/10 patients with the lo-west costs being generated by those who died early.The 10 most expensive patients consisted of mostly younger patients with severe disease who had to be transferred to a nursing home despite maximal efforts of therapy and rehabilitation.Others survived severely impaired and also required costly nursing care.
Our statistical multiple regression analysis confirms that inpatient rehabilitation had the largest impact on mean costs in the first year (p <0.0001).Second, the severity as expressed by the Rankin Score had a linear increasing effect on costs (regression coefficient 0.175, p = 0.0002).

462
A cost analysis of the first year after stroke -early triage and inpatient rehabilitation may reduce long term costs The outcome data of the patients with inpatient rehabilitation are compared with those without.Note the increase of the Barthel-index and the percentage of patients at the nursing home; 81% of the patients with inpatient rehabilitation were independent at home after 1 year.
It is important to note that the costs included total health insurance costs in the first year and therefore also consisted of costs not directly related to the acute stroke.However, on careful examination, most of the costs were related to the acute event and the concomitant vascular risk factors: The mean age of our population was 76 ± 12 years.25% of the patients were diabetics, 31% had suffered TIA or stroke before and 23% had atrial fibrillation.

More versus less expensive patients
A rank order of the 50% more expensive and 50% less expensive patients was established and correlated with the clinical outcome: 23/31 nursing home patients were in the cost-intensive group.As expected, the cost-intensive group usually had a more severe clinical course and only 39% of the patients reached a Barthel-index of 100, whereas the less costly patients reached an index of 100 in 81% (P <0,05).

Independent versus dependent patients
The analysis of independence after 1 year revealed that the costs of the independent patients were lower already in the first year (28,384 ± 20,763 CHF vs. 47,780 ± 26,276 CHF for the dependent patients, p <0.01), despite the fact that approximately half of the independent patients were admitted to the rehabilitation clinic as inpatients.A detailed overview of this subgroup analysis is given in figure 3.
58/131 (44 %) were rehabilitated in the rehabilitation clinic and accumulated 25,908.-CHFper patient with particular treatment, whereas the costs for the nursing home were at 25,799.-CHF per patient in the first year (table 2).Since the subgroup of patients without rehabilitation included twice the number of nursing home patients and 47/58 of the patients with rehabilitation were independent after one year, it appears likely that the initial "costly" inpatient rehabilitation might result in reduced long-term costs.In fact, the patients who underwent the rehabilitation programme increased their Barthel-index by 42 ± 29 points as compared to patients without (inpatient) rehabilitation whose functional level rose by 23 ± 26 points only (p <0.05).This significant difference underlines the efficacy of the patient selection and/or the rehabilitation programme (table 4).

Hospital/acute phase
The acute hospital costs are rather modest (21% of total costs, or 6,403 ± 6,579 CHF per patient with the particular treatment); an additional amount of 1,240 CHF per patient or 19% of the acute hospitalisation costs has to be added, if the costs of the private insurances are included.It also reflects re-hospitalisation in the first year after stroke, eg, occurrence of a second event or need for carotid surgery etc.The median length of stay (LOS) in the acute hospital was 15 days (figure 2 and table 2).

Rehabilitation clinic
The in-patient rehabilitation is most prominent with 37% or a mean of 25,908 ± 14,546 CHF (from 5,544 to 58,806, median 21,504 CHF).Most of these patients (47/58) were independent at the end of the observation period (fig. 1 and tab.2).The LOS in the rehabilitation clinic was 39 days.If the costs of the privately insured patients are added, the mean increases by 3,220 CHF or by 12%.

Long-term nursing home
34% of all patients required long-term nursing home care which is second in the ranking of the cost subgroups, namely 21% or 25,799 ± 19,137 CHF, (min.866, max.79,797, median 24,786 CHF).

Other categories
Smaller contributions to the total costs consisted of outpatient nursing care (1.2%), primary care physician (5%), and medication costs (6.9%).47% of the patients (62/131) were treated by physiotherapists resulting in 3.3% of the total costs (figure 2).Concerning costs per patient of particular treatments, these other categories generated lower costs as compared to the three major categories mentioned above (table 2).

Cost analyses by categories
Stroke outcome data in Switzerland show that 20-30% of the patient still die during the first year [3] and 40-50% of the survivors remain mildly to moderately impaired [4].Survivors of ischaemic strokes have a mortality of 44% over the following five years and 20-40% suffer a second stroke [5].Improved treatment options and the increasing prevalence of stroke may evoke augmented total stroke costs in future.This study correlated the costs after one year of stroke with the clinical outcome of the patients.Our cost analysis in the year after stroke found mean costs per patient of CHF 31'115.-.Stroke management led to a high level of independence after inpatient stroke rehabilitation as a result of the efficiency of patient selection to receive inpatient rehabilitation and/or the effect of inpatient rehabilitation itself.

Annual costs
The results of a Danish stroke study with similar diagnostic and therapeutic modalities showed total costs in the first year of 25,500 $ (= 33,150 CHF, based on a conversion rate of 1,30 at the time of the study).Denmark spent 70% of the costs for the acute hospitalisation and the inpatient rehabilitation [6].In our study, acute hospitalisation and inpatient rehabilitation made up only 58% of the total costs which may be due to careful triage of patients.Long-term costs of stroke in North-America, Australia, New Zealand and several European countries were found to be highly variable depending on which calculation the analysis is based on [7].Nevertheless, the annual costs of 54'546 $ (=70'900.-CHF)were substantially higher than our costs presented.However, it is important to note that in our case only the costs of the insurance were calculated.Thus, it should be kept in mind that the insurance costs underestimate the total costs carried by the state (cantons), the private caregivers, the acute hospitals, and the insurances of the employers.An extrapolated estimate of all these additional costs would be expected to add up to more than double the amount reimbursed by the insurances (73.000.-).This calculation is based on the following assumptions: The state (canton) covers approximately 50% of the costs of the acute hospital, only 23% of nursing home costs, the cost of caregivers at home are 4x higher than those covered, 10% of dependent survivors are 65 years old or less and will lose their working capacity which is conservatively calculated with costs of 50,000.annually.Taken together, the annual costs in our study were comparable to those published for other countries.

Lifetime costs
In the USA costs of stroke ranged from 16.000 to 26,000 $ (= 21,000-34,000 CHF), whereas the lifetime costs were calculated to be 90.000$ (=117,000 CHF) [8].The strokes in the Netherlands contributed to 3,4% of the total health care costs and were at 0,3% of gross domestic product [9].These data illustrate the socioeconomic impact of cerebrovascular disease, particularly since the insurance costs reflect only part of the expenses.This is underlined by nursing home costs in Switzerland [10]: 41% of this sum are paid by the patients, 23% by the health insurance companies, 23% by the state, 9% by the cantons and the communities, and 4% by miscellaneous sources.This substantial contribution of relatives is not reflected by the statistics but can be extrapolated by the fact that women cost 17% more because they are more likely to survive their partners and their home care cannot be provided as frequently as when they themselves care for their spouses.Therefore, the effective nursing home costs would run four times higher than insurance costs reflect.

Hospital/acute phase
The costs of the inpatient phase in our acute hospital lies in third place of the total costs (21%, 6,403 ± 6,579 CHF).In Germany (2000), the first phase (acute hospitalisation) costs 3,840 € (= 5,950 CHF) [11] in the USA (2001) 5,837$ (= 7,580 CHF) [12,13].Another recent large USstudy in 11,000 patients [14] found acute hospital costs of 6,206 $ (= 8,068 CHF).The numbers are surprisingly similar, despite major differences in insurance systems, the variability of the diagnostic workup, therapy, lengths of hospital stay and staff costs.
Our results should be judged cautiously as well since they reflect the lump sum system per case (ie, standard costs per patient hospitalised) and does not necessarily correlate with the severity of the case, the intensity of the diagnostic and therapeutic measures or the length of stay.If the additional costs generated by private insurance are included in our system, the acute inpatient phase costs rise by 1,240 CHF /patient or 19%, and the rehabilitation costs by 3,220 CHF per patient or 12%.We decided not to include these costs in the primary calculations.An interesting alternative in the Swiss health insurance system includes the development of a combined compensation for both the acute hospital and the rehabilitation clinic which would represent a further step in the direction of disease management but would put the acute hospital into the role of the gate keeper with an incentive to reduce the percentage of patients admitted to rehabilitation clinics.

Rehabilitation clinic
Inpatient rehabilitation is the most expensive part of the total cost (37%).However, taking into account the costs per patient with the particular treatment, inpatient rehabilitation is very similar to long-term nursing home after one year already (25,908 CHF vs. 25,799 CHF).The crucial benefit of inpatient rehabilitation is the high percentage of independent patients after one year (81%) which might be associated with reduced longterm costs.
The expenses for the primary physician were strikingly low with 5% of the total costs despite a large number of controls required included blood pressure control, blood sugar, lipids, psychosocial aspects etc.Our analysis suggests that there is still much potential of reducing the costs in strict implementation of preventative strategies for risk factors, in the early admission for stroke treatment, in the further development of stroke teams and of stroke units.The early and well reflected triage decisions for rehabilitation are likely to influence the long-term costs, particularly with a well developed stroke network which spreads over several institutions.
We conclude that the mean costs per patient over one year after stroke generated by our stroke management is similar to those observed in other countries.Patients who die early or who reach independence early are less expensive, already in the first year.Further, patients admitted for inpatient stroke rehabilitation initially generate higher than average costs to insurance companies.However, their Barthel-index significantly increases as compared to the others and 81% of these patients live independently after one year.Therefore, efficient selection of patients for inpatient rehabilitation and the rehabilitation itself might result in lower long-term costs.

A cost analysis of the first year after stroke -early triage and inpatient rehabilitation may reduce long term costs
* The 15 deaths include 13 in-hospital deaths.